REVIEW 2 major objections 3 minor 39 references
Bernoulli determinants and cuspidal subgroups
T0 review · 2 major / 3 minor · reviewed 2026-08-02 · deepseek-v4-flash
Pith's one-line read This paper proves an explicit formula for the order of the rational cuspidal class group of the modular curve X_1(N) for every integer N ≥ 5, built from a determinant of second-Bernoulli-polynomial values.
desk verdict New explicit formula for the rational cuspidal class group of X_1(N) for all N; proof is sound but hinges on a precise reading of Streng's theorem. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The engine of the paper is the Bernoulli transform B_{k,N}: f ↦ Σ_y B_k({xy/N}) f(y). Its determinant on the even (resp. odd) subspace of functions on Z/NZ is 2^{⌊(N−1)/2⌋} D_{k,N}, so evaluating the determinants D_{k,N} reduces to diagonalizing the transform. The transform commutes with the action of (Z/NZ)^×, hence splits into Dirichlet-character eigenspaces V_{N,χ}; on each eigenspace the matrix becomes a product of local matrices over the primes dividing N, whose determinants are computed by induction. On the arithmetic side, the index computation rests on Siegel functions: the divisors of g_{0,a} are read off from B_2({ak/N}), and Streng's theorem describing the full group of modular un
What would settle it
Compute the right-hand side of Theorem 1.1 for a concrete composite N (e.g., N = 12 or N = 15) and compare it with a direct computation of the order of C^Q_1(N) obtained by writing down the divisor matrix of the Siegel units, reducing modulo principal divisors, and taking Galois invariants; a single mismatch would refute the theorem. A cheaper check is to verify Theorem 4.1 numerically for D_{2,11} or D_{3,11} with a computer algebra system, since the cuspidal-group proof depends on that determinant evaluation.
Extended reading notes
Core claim
The central claim, Theorem 1.1, is that for every integer N ≥ 5, |C^Q_1(N)| equals gcd(2,N)^2 · 9/(2^{N−5} N) · ∏_{d|N, d≥3} d^{φ(d)/2} · ∏_{p|N} p^2/(p^2−1) times the product over even primitive Dirichlet characters χ of conductor dividing N of B_{2,χ}^{σ_0(N/Nχ)} (N/Nχ)^{(1/2)σ_0(N/Nχ)} · ∏_{p|N, p∤Nχ} (1−χ(p)p^{−2})^{v_p(N)σ_0(N^{(p)}/Nχ)}. The proof goes by computing the cokernel of the divisor map from the group of modular units to the degree-zero cuspidal divisors: it first expresses the index of the subgroup generated by Siegel units as |D_{2,N}| up to explicit simple factors, and then uses Streng's theorem to compute the index of the full modular-unit group inside that subgroup as 12
Load-bearing premise
The formula's one external load-bearing input is Streng's theorem describing every modular unit on X_1(N) as a unique product of specified Siegel functions whose exponents satisfy two congruences; if that description is inaccurate, the index (div(S):div(U)) = 12 gcd(2,N)N would change and the final formula would fail.
Editorial extensions
If this is right
- For every N ≥ 5, the order of C^Q_1(N) is computable in closed form from Dirichlet L-values and elementary factors; no case-by-case modular computation is needed.
- Theorem 4.1 proves all determinants D_{k,N} are non-zero, giving a uniform proof of independence for the Siegel functions and for the K_2 elements whose non-vanishing was previously checked only for N = 7, 8, 10.
- When N is prime, the general formula reduces to the known expression |C^Q_1(p)| = p^{(p−1)/2} / 2^{p−3} ∏_{χ ≠ 1 even} B_{2,χ}, recovering previous results in one stroke.
- The factorization of the determinant over the primes dividing N yields an Euler-product-style expression for the cuspidal group order, with each prime contributing a local factor.
- The non-vanishing of D_{2,N} combined with Proposition 5.4 gives an independent proof that the Siegel functions g_{0,a} with 1 ≤ a ≤ ⌊N/2⌋ form a Q-basis of U_Q.
Reading between the lines
- The same determinant-eigenspace method could be applied to the full cuspidal subgroup C_1(N) (not just its Galois-invariant part); a formula there would be a major step toward the conjecture that C^Q_1(N) equals the rational torsion of J_1(N), which the paper leaves open.
- The higher-weight cuspidal group introduced in Section 6 can be tested numerically for small N: for an elliptic curve with a rational N-torsion point, the residue map on the tame K_2 group should be controlled by D_{3,N}, so a computer calculation for N = 7 could confirm or refute the speculated link.
- The factorization over primes suggests that the cuspidal class group itself (not just its order) may decompose according to the prime factors of N; the paper does not establish this, but the formula for the determinant is the natural first step toward such a structural theorem.
- Because D_{k,N} is a 'paratrophic' determinant in the sense of Frobenius, the explicit formula likely extends to other congruence subgroups by replacing the character set P^{(-1)^k}_N with the corresponding set; this would give analogous cuspidal formulas for X_0(N) or X(N).
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper gives a closed formula for the order of the rational cuspidal class group C^Q_1(N) of X_1(N) for every N ≥ 5. The proof proceeds in two main steps: first, an independent evaluation of the Bernoulli determinant D_{k,N} (Theorem 4.1), obtained by decomposing the Bernoulli transform into Dirichlet-character eigenspaces and factoring the resulting matrix over prime powers; second, a geometric computation of the two indices in Eq. (15): the index of the Siegel-unit lattice in the rational cusp-divisor group (Proposition 5.4) and the index of the full modular-unit lattice in that Siegel-unit lattice (Proposition 5.6), the latter using Streng's theorem on generators of modular units on X_1(N). The paper also defines a speculative higher-weight analogue of the cuspidal class group and conjecturally relates its order to D_{k,N}.
Significance. If the proof is accepted, Theorem 1.1 is the first uniform formula for the rational cuspidal class group of X_1(N) for composite N, covering and unifying earlier prime-level and twice-prime-level results. The determinant evaluation in Theorem 4.1 is a useful result in its own right, and the paper connects it both to known formulas for cuspidal subgroups and to open questions involving Eisenstein symbols. The argument is largely explicit and checkable: the determinant computation is written out, the N=5 case gives |C^Q_1(5)|=1 as required by genus 0, and the theorem specializes correctly to the known formula for X_1(p). These cross-checks substantially increase confidence in the central claim.
major comments (2)
- [§5.6, Eq. (15)] The factor (div(S):div(U)) = 12 gcd(2,N)N is load-bearing for Theorem 1.1, and its proof rests entirely on the exact form of Streng's theorem [30, Theorem 1.2], including the generator set H_1,...,H_r and the two congruences modulo 12 and gcd(2,N)N. The theorem is not stated in the paper, and the promised direct check for N=4,5 is omitted. Since a mis-transcription of the second modulus or of the generator set would change this index and hence the final formula, please state Streng's theorem explicitly in the form used, and include the N=4,5 check for surjectivity of the map ψ.
- [§5, Lemmas 5.2–5.3 and Prop. 5.4] Proposition 5.4 depends on the parametrization of Galois orbits of cusps (Lemma 5.2) and on the divisor formula for the Siegel units (Lemma 5.3). Both proofs are largely delegated to the author's preprint [6] (Lemmas 5.9 and 5.11), and the exceptional case N=4 in Lemma 5.3 is asserted without calculation. Since these lemmas feed directly into the determinant matrix and the degree computation, an error in any of these statements would propagate to Theorem 1.1. Please either include full proofs or state these as self-contained lemmas with enough detail to verify the exceptional cases.
minor comments (3)
- [§2, Lemma 2.1] The proof says the basis is given by [i]+[-i] for 0≤i≤⌊N/2⌋. For i=0 (and for i=N/2 when N is even) this vector is twice the characteristic function of the corresponding orbit, which affects the coefficient of the row in the displayed matrix. The determinant statement is correct, but the row-scaling discussion should be worded more carefully to avoid confusion.
- [Eq. (16)] The displayed constant in Eq. (16) is difficult to read in the typeset version (the factors involving 2, 3, 5 and the powers of gcd(2,N) are hard to disentangle). Please re-typeset all powers explicitly.
- [§6, Definition 6.1] The higher-weight cuspidal class group is introduced as an image in a higher Chow group, but finiteness is only conjectural and the relation to D_{k,N} is explicitly speculative. The section would be clearer if the statement were labelled as a conjecture rather than presented as an expected theorem.
Circularity Check
No significant circularity; derivation is a genuine computation against external benchmarks (Streng's theorem, classical Siegel-unit divisors); only minor non-load-bearing self-citations.
full rationale
The main formula (Theorem 1.1) is assembled in Eq. (15) as |C^Q_1(N)| = (Div^0(C):div(S)) · (div(S):div(U)). The first factor is computed within the paper in Proposition 5.4: the divisor matrix of the Siegel units is expressed, via the vanishing-order lemma (Lemma 5.3), as a scaled Bernoulli matrix whose determinant is D_{2,N}, and D_{2,N} is evaluated independently in Theorem 4.1 (Sections 2–4) using only the Bernoulli transform, Dirichlet characters, and standard L-value facts. There is no feedback from the cuspidal group to the determinant computation; D_{2,N} is not defined in terms of the target, and Theorem 4.1 is not forced by the index. The second factor, Proposition 5.6, rests on Streng's [30, Theorem 1.2] describing the lattice of modular-unit exponents; Streng is an external, published theorem by a different author, so this is ordinary input-dependence rather than self-citation or a smuggled ansatz. The residual risk flagged in the reader's take — that a mistranscription of the modulus gcd(2,N)N or of the congruence lattice in [30] would change (div(S):div(U)) — is a correctness/transcription risk, not circularity (rule 5). The author's self-citations are auxiliary: [6] supplies Lemmas 5.2 and 5.3 (cusp structure and vanishing orders), but both lemmas are stated in full and sketched in the paper, are parameter-free classical facts, and do not include the target result; [7] is the origin of the D_{3,N} question and is not used in the proof of Theorem 1.1. For N prime the formula reduces to the known formula (2) — a consistency check rather than a renaming, since composite N is genuinely new. No fitted parameters, no prediction forced by construction, and no uniqueness claim imported from the authors' own prior work appear. Overall: no significant circularity; the presence of a minor non-load-bearing self-citation warrants score 2 rather than 0.
Assumptions & free parameters
assumptions (5)
- domain assumption Streng's theorem [30, Theorem 1.2]: the functions H_1,...,H_r generate the group of modular units O(Y_1(N))^×/Q^×, with exponents satisfying sum e_a ≡ 0 mod 12 and sum a^2 e_a ≡ 0 mod N' where N' = gcd(2,N)N.
- domain assumption Divisor formula for Siegel units (Lemma 5.3): ord_{1/k}(g_{0,a}) = N/(2 gcd(k,N)) B_{2,N}(ak) for N ≠ 4, with a special case for N=4; relies on the width of cusps of X_1(N) as in [9, Prop. 6.3.20] and the author's earlier work [6, Lemmas 5.9, 5.11].
- standard math Manin–Drinfeld theorem: the cuspidal subgroup C_1(N) is finite.
- standard math H^1(Gal(Q̄/Q), Q̄^×) = 0 (Hilbert 90).
- standard math Non-vanishing of Dirichlet L-values L(χ,1-k) for k ≥ 2, via the functional equation and Euler product.
invented entities (1)
-
Higher weight rational cuspidal class group C^Q_{k,1}(N) for k ≥ 3
Cite this review
Pith. "Pith review of Bernoulli determinants and cuspidal subgroups." pith.science (2026). https://pith.science/paper/TVYAOGJ3
@misc{pith2026260713536,
author = {Pith},
title = {Pith review of: Bernoulli determinants and cuspidal subgroups},
year = {2026},
howpublished = {\url{https://pith.science/paper/TVYAOGJ3}},
note = {Machine review of arXiv:2607.13536}
}
abstract
We give an explicit formula for the order of the rational cuspidal class group of the modular curve $X_1(N)$ for an arbitrary integer $N$. The proof relies on results of Streng on the group of modular units on $X_1(N)$, and requires computing a certain determinant involving the second Bernoulli polynomial. We also define a higher weight analogue of the cuspidal class group and speculate that its order is related to a similar determinant defined using a higher degree Bernoulli polynomial.
Reference graph
Works this paper leans on
-
[21]
H. Liu , Paratrophic Determinants over Z /N Z via Discrete Fourier Transform, Preprint (2026), arXiv:2603.14795v2 [math.NT] https://arxiv.org/abs/2603.14795v2
arXiv 2026
-
[6]
F. Brunault , On the K_4 group of modular curves, Preprint (2025) arXiv:2009.07614v3 [math.NT] https://arxiv.org/abs/2009.07614v3
arXiv 2025
-
[1]
Arakawa , T
T. Arakawa , T. Ibukiyama and M. Kaneko , Bernoulli numbers and zeta functions. Springer Monogr. Math., Springer, Tokyo, 2014
2014
-
[2]
A. A. Be ilinson , Higher regulators of modular curves, in Applications of algebraic K-theory to algebraic geometry and number theory (Boulder, CO, 1983), Contemp. Math. 55 (Amer. Math. Soc., Providence, RI, 1986), 1--34. doi:10.1090/conm/055.1/862627 https://doi.org/10.1090/conm/055.1/862627
-
[3]
Bertolini , H
M. Bertolini , H. Darmon and K. Prasanna , Generalized Heegner cycles and p -adic Rankin L -series. With an appendix by B. Conrad , Duke Math. J. 162 (2013), no. 6, 1033--1148
2013
-
[4]
Bloch , Algebraic cycles and higher K -theory, Adv
S. Bloch , Algebraic cycles and higher K -theory, Adv. in Math. 61 (1986), no. 3, 267--304. link https://core.ac.uk/download/pdf/82441194.pdf
arXiv 1986
-
[5]
Bloch , The moving lemma for higher Chow groups, J
S. Bloch , The moving lemma for higher Chow groups, J. Algebr. Geom. 3 (1994), no. 3, 537--568
1994
-
[7]
F. Brunault , R. de Jeu , H. Liu and F. Rodriguez Villegas , K_2 of families of elliptic curves over non-Abelian cubic and quartic fields, Preprint (2024), arXiv:2401.04510v1 [math.NT] https://arxiv.org/abs/2401.04510v1
arXiv 2024
Show all 39 references
-
[8]
Cartier , An introduction to zeta functions, in From number theory to physics (Les Houches, 1989), 1--63, Springer-Verlag, Berlin, 1992
P. Cartier , An introduction to zeta functions, in From number theory to physics (Les Houches, 1989), 1--63, Springer-Verlag, Berlin, 1992
1989
-
[9]
Cohen and F
H. Cohen and F. Strömberg , Modular forms. A classical approach. Grad. Stud. Math. 179 (American Mathematical Society, Providence, RI, 2017)
2017
-
[10]
Conrad , The origin of representation theory, Enseign
K. Conrad , The origin of representation theory, Enseign. Math. (2) 44 (1998), no. 3-4, 361--392
1998
-
[11]
Conrad , B
B. Conrad , B. Edixhoven and W. Stein , J_1(p) has connected fibers, Doc. Math. 8 (2003), 331--408
2003
-
[12]
Deligne , Formes modulaires et représentations -adiques, in Sémin
P. Deligne , Formes modulaires et représentations -adiques, in Sémin. Bourbaki 1968/69, no. 355. Lect. Notes Math. 179 (1971), 139--172
1968
-
[13]
Doran and M
C.\,F. Doran and M. Kerr , Algebraic K -theory of toric hypersurfaces, Commun. Number Theory Phys. 5 (2011), 397--600
2011
-
[14]
Drinfeld , Two theorems on modular curves, Funkcional
V.\,G. Drinfeld , Two theorems on modular curves, Funkcional. Anal. i Prilo zen. 7 (1973), no. 2, 83--84
1973
-
[15]
Dummigan , V
N. Dummigan , V. Golyshev , R. de Jeu and M. Kerr , The 2-part of the Bloch-Kato conjecture, and indivisibility results, for K_2 of some elliptic curves, Preprint (2026), arXiv:2605.11100v1 [math.NT] https://arxiv.org/abs/2605.11100v1
2026 arXiv
-
[16]
Gille and T
P. Gille and T. Szamuely , Central simple algebras and Galois cohomology. 2nd revised and updated edition. Cambridge Studies in Advanced Mathematics 165, Cambridge University Press, 2017
2017
-
[17]
Huber , J
T. Huber , J. Opoku and D. Ye , Enumeration of modular forms for _1(N) , Preprint (2026), arXiv:2606.04203v1 [math.NT] https://arxiv.org/abs/2606.04203v1
2026 arXiv
-
[18]
Kato , p -adic Hodge theory and values of zeta functions of modular forms, Ast\'erisque 295 (2004), 117--290
K. Kato , p -adic Hodge theory and values of zeta functions of modular forms, Ast\'erisque 295 (2004), 117--290. https://www.numdam.org/item/AST\_2004\_\_295\_\_117\_0/ https://www.numdam.org/item/AST_2004__295__117_0/
2004
-
[19]
P. E. Klimek , Modular Functions for _1(N) . PhD Thesis, University of California, Berkeley, 1975
1975
-
[20]
D. S. Kubert and S. Lang , Modular units. Grundlehren der Mathematischen Wissenschaften 244, Springer-Verlag, New York-Berlin, 1981
1981
-
[22]
Lupoian , Modular U nits on X_ 1 ( p) and Q uotients of the C uspidal G roup, Preprint (2025), arXiv:2502.04084v1 [math.NT] https://arxiv.org/abs/2502.04084v1
E. Lupoian , Modular U nits on X_ 1 ( p) and Q uotients of the C uspidal G roup, Preprint (2025), arXiv:2502.04084v1 [math.NT] https://arxiv.org/abs/2502.04084v1
2025 arXiv
-
[23]
Mazur , Modular curves and the Eisenstein ideal, Inst
B. Mazur , Modular curves and the Eisenstein ideal, Inst. Hautes Études Sci. Publ. Math. 47 (1977), 33--186
1977
-
[24]
Mestre and N
J.-F. Mestre and N. Schappacher , Séries de Kronecker et fonctions L des puissances symétriques de courbes elliptiques sur , in Arithmetic algebraic geometry (Texel, 1989), 209--245, Progr. Math. 89, Birkhäuser Boston, Inc., Boston, MA, 1991
1989
-
[25]
Ohta , Eisenstein ideals and the rational torsion subgroups of modular Jacobian varieties, J
M. Ohta , Eisenstein ideals and the rational torsion subgroups of modular Jacobian varieties, J. Math. Soc. Japan 65 (2013), no. 3, 733--772
2013
-
[26]
Rosenschon and V
A. Rosenschon and V. Srinivas , Étale motivic cohomology and algebraic cycles, J. Inst. Math. Jussieu 15 (2016), no. 3, 511--537
2016
-
[27]
Schappacher and A
N. Schappacher and A. J. Scholl , The boundary of the Eisenstein symbol, Math. Ann. 290 (1991), no. 2, 303--321. Erratum in vol. 290 (1991), no. 4, p. 815
1991
-
[28]
A. J. Scholl , Motives for modular forms, Invent. Math. 100 (1990), no. 2, 419--430
1990
-
[29]
Steinberg , Factoring the Dedekind-Frobenius determinant of a semigroup, J
B. Steinberg , Factoring the Dedekind-Frobenius determinant of a semigroup, J. Algebra 605 (2022), 1--36
2022
-
[30]
Streng , Generators of the group of modular units for ^1(N) over the rationals, Ann
M. Streng , Generators of the group of modular units for ^1(N) over the rationals, Ann. Henri Lebesgue 6 (2023), 95--116
2023
-
[31]
Takagi , Cuspidal class number formula for the modular curves X_1(p) , J
T. Takagi , Cuspidal class number formula for the modular curves X_1(p) , J. Algebra 151 (1992), no. 2, 348--374
1992
-
[32]
Takagi , The cuspidal class number formula for the modular curves X_1(p^m) , J
T. Takagi , The cuspidal class number formula for the modular curves X_1(p^m) , J. Algebra 158 (1993), no. 2, 515--549
1993
-
[33]
Takagi , The cuspidal class number formula for the modular curves X_1(3^m) , J
T. Takagi , The cuspidal class number formula for the modular curves X_1(3^m) , J. Math. Soc. Japan 47 (1995), no. 4, 671--686
1995
-
[34]
Takagi , The -rational cuspidal group of J_1(2p) , J
T. Takagi , The -rational cuspidal group of J_1(2p) , J. Math. Soc. Japan 66 (2014), no. 4, 1249--1301
2014
-
[35]
Yang , Modular units and cuspidal divisor class groups of X_1(N) , J
Y. Yang , Modular units and cuspidal divisor class groups of X_1(N) , J. Algebra 322, no. 2, 514--553 (2009)
2009
-
[36]
Yang and J-D
Y. Yang and J-D. Yu , Structure of the cuspidal rational torsion subgroup of J_1(p^n) , J. Lond. Math. Soc., II. Ser. 82, no. 1, 203--228 (2010)
2010
-
[37]
Yoo , The rational cuspidal divisor class group of X_0(N) , J
H. Yoo , The rational cuspidal divisor class group of X_0(N) , J. Number Theory 242 (2023), 278--401
2023
-
[38]
Yoo , The rational torsion subgroup of J_0(N) , Adv
H. Yoo , The rational torsion subgroup of J_0(N) , Adv. Math. 426 (2023), Paper no. 109111
2023
-
[39]
Yu , A cuspidal class number formula for the modular curves X_1(N) , Math
J. Yu , A cuspidal class number formula for the modular curves X_1(N) , Math. Ann. 252 (1980), no. 3, 197--216
1980
Reviewed August 2, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.